S01 · E01 The Pilot
Kids…
how i met your AI Engineer. Data Analyst. BI Developer. Data Scientist.

Spoiler alert: that's me. I build AI products, automate workflows, and turn messy data into decisions worth telling stories about. Everything from SQL dashboards to LangChain agents. But every good story starts at the beginning...

Start the story →
Paritosh Vyawahare
Paritosh · Boston '26
STARRING Paritosh AI · DATA · BI
"

Data is the new story. And I'm here to tell the right ones.

T. MOSBY
📍 Boston, MA · Open to full-time roles
Paritosh Vyawahare
Paritosh · Boston '26
For Robin. ♥
The Beginning

Kids, every good story has a beginning. Mine starts back in 2022, in India. Two years of data work. Dashboards at Larsen & Toubro. Machine learning pipelines at Aspen Systems. Then in 2024, I packed up my life and moved to Boston for a Master's in Data Analytics Engineering at Northeastern University. Finished it in April 2026.

Now somewhere in the middle of all that, something clicked. I didn't stop being the person who makes the dashboard. I just added another skill on top of it. Now I build the dashboard AND the AI agent that can read it, question it, and act on it. That's why I work across the whole data-to-decision pipeline. SQL, Power BI, and Tableau on one end. LangChain, RAG, and Groq on the other. Same story. Bigger toolkit.

So that brings us to right now. I'm looking for a full-time role in AI Engineering, Data Science, Data Analytics, or Business Intelligence. If any of those sound like something you're building, well. Let's talk.

True story.
The Timeline, In Icons
2019
Computer Science Engineering
2022
Data Analyst · L&T
2023
Data Scientist · Aspen
2025
Data Analytics · Northeastern
Today
Building AI Products

but that's a story for another time...

✦  EP. 508  ·  THE PLAYBOOK  ✦

The Playbook.

Every good story has its plays. Here are five of mine. Across agentic AI, analytics, ML, and BI. Each one solved a real problem. Each one shipped something you can click.

Play / 01
Agentic AI · Data Analyst
LATEST

Vera.AI

The AI data analyst that only speaks in numbers.

This one's my favorite. You upload a CSV, and Vera hands back a full statistical profile, an AI-written analyst report, auto-generated Plotly charts, and a chat where you can ask her anything. Four coordinated agents under the hood: Analyzer, Report, Chart, and Q&A. She writes pandas code in a security-hardened sandbox, runs it, and shows you the exact code that ran. Every claim traceable to your data. No hallucinations. No made-up numbers.

PythonStreamlitLangChainGroq · Llama 3.3 70BPandasscikit-learnPlotlyDockerHugging Face
Try Vera live → Watch walkthrough → GitHub →
Play / 02
Data Eng · BI

Policy Lapsation

End-to-end ETL and BI for life insurance lapse analysis. 90K policies, 6+ KPIs.

Talend pipelines into a PostgreSQL warehouse, dashboards in Power BI and Tableau. Surfaces lapse rate by policy age, premium band, demographics, and geography, designed for the retention team to act on, not just look at.

TalendPostgreSQLPower BITableauSQLETL
GitHub →
Play / 03
Agentic AI

MeetingFlow

An AI agent that turns meeting transcripts into Notion tasks and follow-up emails.

Multimodal input (text or Whisper audio), Groq's Llama 3.3 70B for structured task extraction, tone-adaptive email drafts based on inferred meeting type, and human-in-the-loop approval before anything hits your workspace. The full agentic loop, built end-to-end. Barney would have delegated this in a heartbeat.

PythonLangChainGroqWhisperNotion APIStreamlit
GitHub →
Play / 04
Machine Learning

Churn Prediction

A predictive model that identifies at-risk customers and drives retention.

Random Forest classifier on 60K customer records, hitting 86% accuracy after feature engineering, hyperparameter tuning, and cross-validation. Identified the real churn drivers: month-to-month contracts and mailed-check payments. Predicted who'd leave. Like Ted, but useful.

Pythonscikit-learnRandom ForestpandasFeature Engineering
GitHub →
Play / 05
Production RAG

SupportAI

A production RAG chatbot with 100% accuracy on its evaluation benchmark.

Retrieval-augmented Q&A over company docs. Multi-query retrieval for better recall, source citations on every answer, conversation memory for natural follow-ups, and refusal guardrails when the docs don't have the answer. Fully containerized. Marshall-level thorough.

PythonFastAPILangChainChromaDBGroqDocker
GitHub →

Nothing good happens after 2 AM.
Except maybe deploying to prod.

S01 · E03  ·  THE STACK

Marshall's Graphs.

If Marshall taught us anything, it's that any answer sounds smarter with a pie chart behind it. So here's the stack, quantified.

How I Spend My Time
based on 100% of a made-up week
LLM & RAG 35% Python / SQL 30% BI & Dash 20% ML 15%
Comfort by Tool
on a scale of "used it once" to "I dream in it"
Python SQL LangChain Power BI PyTorch high meh low ← more time spent

Languages

  • Python
  • SQL
  • R

AI & ML

  • LangChain
  • LLMs (Groq, OpenAI)
  • RAG & Vector DBs
  • PyTorch
  • scikit-learn

Analytics & BI

  • Power BI
  • Tableau
  • Excel
  • Pandas

Data & Cloud

  • PostgreSQL
  • MongoDB
  • AWS
  • Azure Data Factory
  • Streamlit
And next? Well kids, legen... wait for it ...dary things:
LangGraph, agent orchestration, and LLM evaluation frameworks.
Suit Up
Barney would approve.

The Journey.

Every good story has a suit-up montage. This is mine. Three roles. Two countries. One very long flight to Boston.

2025 to 2026
Boston, MA
Data Analytics Engineer
Northeastern University Research Enterprise Services
Built interactive Tableau and Power BI dashboards for research operations and financial analytics. Automated a complex reporting process in Python that used to eat a full morning every week. Set up Tableau Prep Builder and Power Automate to streamline the data prep and refresh cycles. Designed advanced Excel models with PivotTables and dynamic formulas to answer executive-level queries.
2023 to 2024
India
Data Scientist
Aspen Systems
And now, kids, we get to the ML part. Built end-to-end machine learning pipelines for client business problems using Python, scikit-learn, and XGBoost. Feature engineering through deployment. Worked across MLOps tooling like Docker, AWS, and CI/CD, to operationalize models reliably. Collaborated with product and engineering teams to ship analytics features to production.
2022 to 2023
India
Data Analyst
Larsen & Toubro
Analyzed large-scale product-level sales data with SQL to uncover geographic trends and underperforming regions. Built interactive Power BI dashboards with KPIs and supporting visuals for operations and executive teams. Owned the analytics layer for several internal initiatives from requirement gathering to delivery.
The Next Chapter
Loading… (that's where you come in.)
Have you met Paritosh?

So that brings us to the end of the story. Well, actually the beginning of the next one. I'm open to full-time roles in AI Engineering, Data Science, Analytics, and BI. Say hi. First round's on me.

LinkedIn GitHub Email me
P.S. True story.
Let's build something legendary together. P.
Challenge
Accepted.